Package org.apache.drill.exec.record.metadata

Interface ColumnMetadata

All Superinterfaces:
Propertied
All Known Implementing Classes:
AbstractColumnMetadata, AbstractMapColumnMetadata, DictColumnMetadata, DynamicColumn, MapColumnMetadata, PrimitiveColumnMetadata, ProjectedColumn, RepeatedListColumnMetadata, VariantColumnMetadata
public interface ColumnMetadata extends Propertied
Metadata description of a column including names, types and structure information.

  • Field Details

    • EXPECTED_CARDINALITY_PROP

      static final String EXPECTED_CARDINALITY_PROP
      Predicted number of elements per array entry. Default is taken from the often hard-coded value of 10.
      See Also:

    • DEFAULT_VALUE_PROP

      static final String DEFAULT_VALUE_PROP
      Default value represented as a string.
      See Also:

    • EXPECTED_WIDTH_PROP

      static final String EXPECTED_WIDTH_PROP
      Expected (average) width for variable-width columns.
      See Also:

    • FORMAT_PROP

      static final String FORMAT_PROP
      Optional format to use when converting to/from string values.
      See Also:

    • BLANK_AS_PROP

      static final String BLANK_AS_PROP
      Indicates how to handle blanks. Must be one of the valid values defined in AbstractConvertFromString. Normally set on the converter by the plugin rather than by the user in the schema.
      See Also:

    • BLANK_AS_NULL

      static final String BLANK_AS_NULL
      Convert blanks to null values (if the column is nullable), or fill with the default value (non-nullable.)
      See Also:

    • BLANK_AS_ZERO

      static final String BLANK_AS_ZERO
      Convert blanks for numeric fields to 0. For non-numeric fields, convert to null (for nullable) or the default value (for non-nullable). Works best if non-numeric fields are declared as nullable.
      See Also:

    • EXCLUDE_FROM_WILDCARD

      static final String EXCLUDE_FROM_WILDCARD
      Indicates whether to project the column in a wildcard (*) query. Special columns may be excluded from projection. Certain "special" columns may be available only when explicitly requested. For example, the log reader has a "_raw" column which includes the entire input line before parsing. This column can be requested explicitly:
      SELECT foo, bar, _raw FROM ...
      but the column will not be included when using the wildcard:
      SELECT * FROM ...

      Marking a column (either in the provided schema or the reader schema) will prevent that column from appearing in a wildcard expansion.

      See Also:

    • DEFAULT_ARRAY_SIZE

      static final int DEFAULT_ARRAY_SIZE
      See Also:

    • IMPLICIT_COL_TYPE

      static final String IMPLICIT_COL_TYPE
      Indicates that a provided schema column is an implicit column (one defined by Drill rather than the reader.) Allows the implicit schema to reify partition names, say, as reader-specific names. For example, dir0 might be reified as year, etc.

      Available when the underlying reader supports implicit columns. The value is the defined implicit column name (not the name set via system/session options.) Using the defined name makes the provided schema immune from runtime changes to column names.

      As the result of adding this feature, any column not tagged as implicit is a reader column, even if that column happens to have the same (currently selected runtime) name as an implicit column.

      See Also:

    • IMPLICIT_FQN

      static final String IMPLICIT_FQN
      Fully-qualified name implicit column type.
      See Also:

    • IMPLICIT_FILEPATH

      static final String IMPLICIT_FILEPATH
      File path implicit column type.
      See Also:

    • IMPLICIT_FILENAME

      static final String IMPLICIT_FILENAME
      File name implicit column type.
      See Also:

    • IMPLICIT_SUFFIX

      static final String IMPLICIT_SUFFIX
      File suffix implicit column type.
      See Also:

    • IMPLICIT_PARTITION_PREFIX

      static final String IMPLICIT_PARTITION_PREFIX
      Prefix for partition directories. dir0 is the table root folder, dir1 the first subdirectory, and so on. Directories that don't exist in the actual file path take a NULL value.
      See Also:

  • Method Details

    • structureType

      ColumnMetadata.StructureType structureType()

    • tupleSchema

      TupleMetadata tupleSchema()
      Schema for TUPLE columns.
      Returns:
      the tuple schema

    • variantSchema

      VariantMetadata variantSchema()
      Schema for VARIANT columns.
      Returns:
      the variant schema

    • childSchema

      ColumnMetadata childSchema()
      Schema of inner dimension for MULTI_ARRAY columns. If an array is 3D, the outer column represents all 3 dimensions. outer.childSchema() gives another MULTI_ARRAY for the inner 2D array. outer.childSchema().childSchema() gives a column of some other type (but repeated) for the 1D array.

      Sorry for the mess, but it is how the code works and we are not in a position to revisit data type fundamentals.

      Returns:
      the description of the (n-1) st dimension.

    • schema

      MaterializedField schema()

    • emptySchema

      MaterializedField emptySchema()

    • name

      String name()

    • type

      TypeProtos.MinorType type()

    • majorType

      TypeProtos.MajorType majorType()

    • mode

      TypeProtos.DataMode mode()

    • dimensions

      int dimensions()

    • isNullable

      boolean isNullable()

    • isArray

      boolean isArray()

    • isVariableWidth

      boolean isVariableWidth()

    • isMap

      boolean isMap()

    • isVariant

      boolean isVariant()

    • isDict

      boolean isDict()

    • isScalar

      boolean isScalar()

    • isDynamic

      boolean isDynamic()
      Reports if the column is dynamic. A dynamic column is one with a "type to be named later." It is valid for describing a dynamic schema, but not for creating vectors; to create a vector the column must be resolved to a concrete type. The context should make it clear if any columns can be dynamic.
      Returns:
      true if the column does not yet have a concrete type, false if the column type is concrete

    • isMultiList

      boolean isMultiList()
      Determine if the schema represents a column with a LIST type with UNION elements. (Lists can be of a single type (with nullable elements) or can be of unions.)
      Returns:
      true if the column is of type LIST of UNIONs

    • isEquivalent

      boolean isEquivalent(ColumnMetadata other)
      Report whether one column is equivalent to another. Columns are equivalent if they have the same name, type and structure (ignoring internal structure such as properties.)

    • setExpectedWidth

      void setExpectedWidth(int width)
      For variable-width columns, specify the expected column width to be used when allocating a new vector. Does nothing for fixed-width columns.
      Parameters:
      width - the expected column width

    • expectedWidth

      int expectedWidth()
      Get the expected width for a column. This is the actual width for fixed- width columns, the specified width (defaulting to 50) for variable-width columns.
      Returns:
      the expected column width of the each data value. Does not include "overhead" space such as for the null-value vector or offset vector

    • setExpectedElementCount

      void setExpectedElementCount(int childCount)
      For an array column, specify the expected average array cardinality. Ignored for non-array columns. Used when allocating new vectors.
      Parameters:
      childCount - the expected average array cardinality. Defaults to 1 for non-array columns, 10 for array columns

    • expectedElementCount

      int expectedElementCount()
      Returns the expected array cardinality for array columns, or 1 for non-array columns.
      Returns:
      the expected value cardinality per value (per-row for top-level columns, per array element for arrays within lists)

    • setFormat

      void setFormat(String value)

    • format

      String format()

    • dateTimeFormatter

      DateTimeFormatter dateTimeFormatter()
      Returns the formatter to use for date/time values. Only valid for date/time columns.
      Returns:

    • setDefaultValue

      void setDefaultValue(String value)
      Sets the default value property using the string-encoded form of the value. The default value is used for filling a vector when no real data is available.
      Parameters:
      value - the default value in String representation

    • defaultValue

      String defaultValue()
      Returns the default value for this column in String literal representation.
      Returns:
      the default value in String literal representation, or null if no default value has been set

    • decodeDefaultValue

      Object decodeDefaultValue()
      Returns the default value decoded into object form. This is the same as: 
      decodeValue(defaultValue());
      Returns:
      the default value decode as an object that can be passed to the ColumnWriter#setObject() method.

    • valueToString

      String valueToString(Object value)

    • valueFromString

      Object valueFromString(String value)

    • cloneEmpty

      ColumnMetadata cloneEmpty()
      Create an empty version of this column. If the column is a scalar, produces a simple copy. If a map, produces a clone without child columns.
      Returns:
      empty clone of this column

    • precision

      int precision()

    • scale

      int scale()

    • bind

      void bind(TupleMetadata parentTuple)

    • copy

      ColumnMetadata copy()

    • typeString

      String typeString()
      Converts type metadata into string representation accepted by the table schema parser.
      Returns:
      type metadata string representation

    • columnString

      String columnString()
      Converts column metadata into string representation accepted by the table schema parser.
      Returns:
      column metadata string representation